摘要： 考虑到信号波动会对无线定位产生影响,基于深度神经网络提出一种回归的无线定位方法。采用四层深度神经网络结构进行定位,通过堆叠去噪自编码器对网络结构进行预训练,避免采用人工设计的方式,从大量有噪的样本中,自动学习有效特征。分不同时段从现实场景中采集数据进行实验,结果表明,针对波动的无线信号,该方法能有效提高定位准确率。

关键词: 无线定位, 深度神经网络, 回归, 深度学习, 堆叠去噪自编码器

Abstract: To deal with the variant and unpredictable wireless signals,this paper proposes a regressive wireless positioning method based on Deep Neural Network(DNN).It uses a four-layer Deep Neural Network(DNN) structure pre-trained by Stacked Denoising Autoencoder(SDA).This network structure is capable of learning reliable features from a large number of noisy samples,avoiding hand-engineering.The data required for the experiments is collected from the real world in different periods.Experimental results indicate that the proposed method can improve positioning accuracy in coping with the turbulent wireless signals.

Key words: wireless positioning, Deep Neural Network(DNN), regression, deep learning, Stacked Denoising Autoencoder(SDA)

中图分类号: 

• TP393